In development of innovative wind turbines as part of the Innwind.EU project, the present

work aimed at identifying possible shortcomings in aeroelastic modeling of Large Wind

Turbines with respect to:

1. Structural modeling of complex designs (geometrical nonlinearities / coupled modes)

2. The increased excitation due to the lowering of the rotor frequencies towards the more

energetic part of the wind spectrum.

3. The increased azimuthal variation of the wind inflow excitation and the related dynamic

inflow modeling

Aims and Content

Cross comparison of aeroelastic state-of-the-art

Tools on a 10MW scale wind turbine

D.I. Manolas1, G.R. Pirrung2, A. Croce3, M. Roura4, V.A. Riziotis1, H.A. Madsen2, C. Pizarro4, S.G. Voutsinas1, F. Rasmussen2 1NTUA, 2DTU, 3PoliMi, 4GAMESA

Fully coupled simulations in Turbulent inflow

Conclusions

EWEA 2015 – Paris – 17-20 November 2015

PO.263

QR

Code (leave blank)

Third edgewise mode

RFC of the flapwise moment – 8m/s PSD of the flapwise moment – 8m/s

Fair agreement with 3D FEM up to the 7th mode

Good agreement (max uncertainty in the torsion

angle=0.4o at the tip)

The bending-torsion geometric nonlinear effect is

triggered due to the high flapwise deflections

(about 10% of the blade length).

In closed loop operation the agreement in the flapwise bending PSD is slightly better. This is

due to the variable speed operation. Then with respect to the edgewise direction, there is

very good agreement in the corresponding PSD of the deflection at tip, at 1P (0.2Hz) and at

the 1st edgewise mode (peak about 0.95Hz). At higher frequencies the response in the

range 1.5-2Hz is predicted differently. In the results of both versions of hGAST the same

frequency for the peak is predicted so it should be attributed to the different modeling of the

nonlinear structural couplings.

Simulations within the INNWIND.EU project from 4 aeroelastic tools and their variants for the

10MW Reference Wind Turbine designed by DTU are compared. In order to address one by

one the different aspects of modeling the following step were taken:

1. For the structural part, modal analysis and static loading cases; comparison with 3D FEM

2. For the assessment of the dynamic inflow, Stiff rotor operation in turbulent wind

3. For the full aeroelastic case: Normal operation in open and closed loop conditions

Structural Cross Comparison

Code name Institute Structural modeling Aerodynamics

Cp-Lambda PoliMi FEM Geometrically Exact FNL / Multi Body BEM

HAWC2 DTU FEM Timoshenko / Multi Body BEM, Near Wake

hGAST NTUA FEM Timoshenko / Multi Body BEM, Free Wake

NEREA GAMESA FEM Generalized Timoshenko BEM

The tools

Modal Analysis

Static flapwise Loading

Twist angle

Aerodynamic simulations

Baseline BEM

vs Unsteady airfoil aerodynamics (uaam)

Baseline BEM

Vs Unsteady airfoil aerodynamics (uaam)

+Dynamic Inflow modeling (dim)

Vs

Near Wake (NW) & Free Wake (GenUVP)

In these simulations the rotor is stiff while the inflow turbulent. When the dynamic inflow is

switched off the ranges in the RFCs agree well. On the contrary when it is switched on the

expected range decrease is not the same. The cylindrical model in hGAST-BEM over-filters

the excitations compared to HAWC2 BEM while the near and the free wake models fall in

between.

Open loop operation

Closed loop operation

• In purely structural terms the beam models was found to compare well to the 3D FEM data up to the 7th mode. 3D FEM predicted in general higher frequencies. Strong coupling effects are

consistently predicted and a fair overall agreement is obtained in the shape of the modes. The beam models under predict the torsion angle and the coupled flapwise mode.

• In terms of unsteady airfoil dynamics, the existing models work similarly. In terms of dynamic wake modeling, the simple cylindrical model over-filters the response compared to more elaborate

ones, while free wake results are in between. As a result in the fully coupled cases, in open loop operation deviations are seen in the 1P & 2P flapwise responses.

• In closed loop operation the agreement in the flapwise predictions slightly improves because of the controller, while different modeling of the bending torsion coupling leads to different response

in the 2nd symmetric edgewise mode.

PSD of the flapwise moment – 8m/s PSD of the edgewise deflection – 8m/s

As in the stiff rotor case the cylindrical wake model under-predicts the ranges compared to the

more elaborate models while the free-wake models fall in between. In the corresponding PSD

the deviations are mainly concentrated on the 1P & 2P response which is also attributed to the

differences in the modeling of the dynamic inflow.

The PSD plot of the pitching moment at

blade root shows good agreement in the

dominant 1P peak which dominates the

response. There is an excitation at ~0.9 Hz

that corresponds to the 1st edgewise mode.

This cross talking is due to a bending-

torsion coupling effect which gets more

pronounced when high flapwise deflections

appear.

PSD of the pitching moment – 8m/s

RFC of Thrust– 8m/s RFC of Thrust– 8m/s

Similar mode shapes - Coupled modes